Many cell surface proteins, including growth factor receptors, are internalized from the cell surface into membrane compartments called early endosomes. The decline in pH in early endosomes causes dissociation of the receptor-ligand complex. Subsequently, the majority of receptors are further directed to late endosomes, often referred to as multivesicular bodies, and delivered to lysosomes for degradation, while others are recycled to the plasma membrane (Katzman et al.,2002; Raiborg et al.,2003). The sorting process within the endosomal system is complex, and the signal for degradation is provided by ubiquitin (Haglund et al.,2003; Huang et al.,2006). Receptors are committed to the lysosomal pathway through four separate protein complexes, called endosomal sorting complex required for transport (ESCRT)-0, -I, -II, and -III. The ubiquitinated endosomal cargo is first recognized and bound by ESCRT-0 (Conibear,2002; Hurley and Emr,2006; Williams and Urbé,2007), which consists of hepatocyte growth factor-regulated tyrosine kinase substrate (HRS) and two signal transducing adaptor molecule (STAM) isoforms, signal transducing adaptor molecule 1 (STAM1) and signal transducing adaptor molecule 2 (STAM2) (Asao et al.,1997; Endo et al.,2000). STAMs have previously been identified as proteins that are highly phosphorylated on tyrosine residues in response to cytokine and growth factor stimulation and are therefore thought to be involved in intracellular signaling as well (Takeshita et al.,1997). They are considered to link membrane transport with intracellular signaling and have a characteristic functional domain organization: Vps-27/HRS/STAM (VHS), ubiquitin-interacting motif (UIM), Src homology (SH3), coiled coil (CC), STAM-specific motif (SSM), and immunoreceptor tyrosine-based activation motif (ITAM) domain (Endo et al.,2000; Kato et al.,2000; Lohi and Lehto,2001; Mizuno et al.,2003,2004). Double knockout of both Stam1 and Stam2 is embryonically lethal, while either Stam1 or Stam2 knockout mice show no abnormalities at birth. Nevertheless, Stam1 knockout adult mice were shown to display loss of neurons in the CA3 region of the hippocampus (Yamada et al.,2002). In general, mutations in components of the ESCRT machinery have recently been linked to many diseases, in particular neurodegenerative diseases and cancer (Stuffers et al.,2009).
Although our initial reverse transcriptase-polymerase chain reaction (RT-PCR) analysis of Stam2 mRNA indicated that Stam2 mRNA is ubiquitously expressed (Curlin et al.,2006), as such confirming the Northern blotting data of Takata et al. (2000), subsequent fine analysis obtained through gene trap mice carrying lacZ reporter in frame with Stam2 specifically revealed strong Stam2 expression in the enteric nervous system. This somewhat unexpected finding led us to perform this study. A variety of signaling pathways and molecules have proven necessary for the normal development and function of the enteric nervous system. Moreover, the unknown significance of the cell signaling and endosome sorting for the maintenance of a healthy gut was another important reason to study Stam2 in the enteric nervous system.
We demonstrated high Stam2 expression in the enteric nervous system, in neurons and in interstitial cells of Cajal (ICCs). Still, the gene trap mutation of Stam2 did not cause phenotype changes of the enteric nervous system.
MATERIALS AND METHODS
Experiments were performed on 4- to 8-month-old male mice from the gene trap mouse line Stam2Gt1Gaj and the wild type inbred strain C57Bl/6NCrl.
The transgenic mouse line Stam2Gt1Gaj was generated by the gene trap method (Skarnes et al.,1992; Thomas et al.,2000). Embryonic stem cells were modified with a nonhomologous DNA vector pKC199βgeo containing a splice acceptor sequence from the mouse Hoxc9 gene and fused promoterless lacZ and neoR genes. The obtained insertion was in the Stam2 gene, in the intron between exons 2 and 3 (Curlin et al.,2006). The Stam2Gt1Gaj transgenic mice were kept as heterozygous on a C57Bl/6NCrl genetic background through 24 generations. Homozygous animals were obtained by intercrossing of heterozygous. Transgenic mice were genotyped by PCR of tail genomic DNA.
All experiments were approved by the institutional Ethical Committee and were in agreement with the Croatian Society for Laboratory Animal Science and the International Council for Laboratory Animal Science.
Detection of β-Galactosidase Activity
The mice were anesthetized with intraperitoneal injections of Avertin (0.5 g/kg) and then perfused transcardially sequentially with phosphate-buffered saline (PBS) and fixative containing 2% formaldehyde and 0.2% glutaraldehyde (Sigma-Aldrich) in 0.01 M PBS (pH 7.4). Esophagus, stomach, duodenum, and distal colon were dissected and immersion fixed with the same fixative for 30 min on ice. After rinsing in PBS, 3–4 cm-long tissue slabs were incubated in the staining solution, containing 0.5 mg/mL X-gal (5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside), 10 mM potassium ferricyanide, 10 mM potassium ferrocyanide, 2 mM magnesium chloride, 0.01% sodium deoxycholate, and 0.01% Igepal (Sigma-Aldrich) in 0.01 M PBS (pH 7.4) under light protection at 37°C overnight. The specimens were rinsed with PBS and cleared in ascending concentrations of glycerol in PBS at 4°C. One part of the tissue was visualized under the stereomicroscope (Olympus SZH10) as a whole mount of the digestive tube; the other part was embedded in paraffin and 20 μm-thin histological sections were prepared and counterstained with nuclear fast red (Chroma-Gesellschaft Schmid Gmbh & Co.).
Unlike the wholemounts of the total digestive wall, those of the muscle layer were prepared by separating the muscle layer from the other layers. These animals were not perfused, but the muscle layer was isolated from unfixed tissue, subsequently fixed by immersion and processed as above.
For conventional histology, animals were fixed by perfusion in 4% paraformaldehyde. The selected parts of the gastrointestinal tract were isolated and fixed further by immersion in the same fixative, for 1 hr on ice. Esophagus, stomach, duodenum, and distal colon were embedded in paraffin and 7 μm-thin sections were prepared and stained with hematoxylin–eosin.
Segments of stomach, duodenum, and distal colon used for immunohistochemical analysis were collected after fixation by perfusion in 4% paraformaldehyde followed by immersion in the same fixative for 3 hr on ice. Each segment was rinsed in PBS, and then transferred to 10% sucrose followed by 30% sucrose in PBS at 4°C. A total of 20 μm-thin frozen sections were cut on a cryostat and immunolabeled on SuperFrost microscope slides (Menzel-Glaser) with primary antibodies against: STAM2 (rabbit polyclonal, diluted 1:150, Santa Cruz Biotechnology, sc-98681), STAM1 (goat polyclonal, diluted 1:100, Santa Cruz Biotechnology, sc-6919), HRS (mouse monoclonal, diluted 1:100, Santa Cruz Biotechnology, sc-166843), microtubule-associated protein 2 (MAP2; chicken polyclonal, diluted 1:700, Abcam, ab5392), protein gene product 9.5 (PGP9.5; mouse monoclonal, diluted 1:20, Abcam, ab8189), glial fibrillary acidic protein (GFAP; chicken polyclonal, diluted 1:50, Abcam, ab4674), c-kit (goat polyclonal, diluted 1:200, Santa Cruz Biotechnology, sc-1494), β-galactosidase (chicken polyclonal, diluted 1:200, Abcam, ab9361), and early endosome antigen 1 (EEA1; mouse monoclonal, diluted 1:500, BD Transduction Laboratories, 610457). Incubation with primary antibodies was performed at 4°C overnight. The secondary antibodies were: Alexa Fluor 488 goat anti-rabbit (Invitrogen, A11008), Alexa Fluor 488 donkey anti-rabbit (Invitrogen, A11056), Alexa Fluor 546 goat anti-chicken (Invitrogen, A11040), Alexa Fluor 546 goat anti-mouse (Invitrogen, A11003), and Alexa Fluor 546 donkey anti-goat (Invitrogen, A21206). All secondary antibodies were diluted at a concentration 1:500, and incubation was performed for 2.5 hr at room temperature. After final rinsing, sections were coverslipped in Fluoromount (Sigma-Aldrich) and analyzed by confocal microscopy (Zeiss LSM 510 Meta).
Quantitative Morphological Analysis
Whole-mount preparations of the muscle layer (described above; prepared by separating the muscle layer from the other layers) stained with X-gal solution were used to estimate the volume density of the X-gal-labeled structures (Van Ginneken et al.,2002). Samples of the muscle layer (3 cm2) were taken from two animals per genotype (homozygous and heterozygous). In each sample, 10 fields, each of 1.3 mm2 were randomly chosen and analyzed. Volume densities were calculated by dividing the number of grid points that overlapped the space of interest (P(I), X-gal stained area) by the number of grid points hitting the reference space (P(ref), muscle layer) of 1.3 mm2. The density of the stereological grid (number of points), the number of sections, and the number of sample fields were chosen to give a coefficient of error (CE) of the estimation that was less than 0.1 (Gundersen and Jensen,1987).
Samples of distal colon (1 cm each) were taken from two animals per genotype (homozygous and wild type) and seven randomly chosen cryosections per each animal were immunostained and used for analysis. The density of the stereological grid, the number of sections, and the number of sample fields were chosen to give a coefficient of error (CE) of the estimation that was less than 0.1 (Gundersen and Jensen,1987). Volume density of the immunoreactive MAP2 areas in the 0.2 mm2 reference space of muscle layer was calculated by dividing the number of grid points hitting the space of interest (P(I), MAP2 positive area) by the number of grid points hitting the reference space (P(ref), muscle layer), as previously described (Van Ginneken et al.,2002).
All measurements of the volume densities of the X-gal-stained areas and MAP2-positive areas from both genotypes (i.e., homozygous and heterozygous, and homozygous and wild type, respectively) were used to test the normality of sample distribution by the Kolmogorov–Smirnov test and the Shapiro–Wilk test. The samples which followed normal distribution (X-gal-stained areas) were statistically analyzed by the Student t test (P < 0.1) and samples, which did not follow normal distribution (MAP2-positive areas) were analyzed by nonparametric Mann–Whitney U test (P < 0.1). Analyses of variance (ANOVA) or Kruskal–Wallis ANOVA were used to compare the means of single animals to confirm the equality of the variances calculated by the comparison of two genotype groups.
β-Galactosidase Activity is Present in the Enteric Nervous System of Stam2Gt1Gaj Mice
Visualization of Stam2 expression in the enteric nervous system was achieved with the genetically modified mouse line generated by the gene trap screen. As a consequence of the gene trap modification, lacZ gene was in frame with Stam2 gene, allowing us to assess Stam2 expression by histochemical detection of β-galactosidase activity via its substrate X-gal, which formed a blue precipitate.
X-gal staining was analyzed on whole-mount fragments of the digestive tube and on cryosections. β-galactosidase activity was present in the myenteric and submucosal plexus of both heterozygous and homozygous animals at all levels of the digestive tract (Fig. 1A–F). Only in the esophagus, where the submucosal plexus is lacking in mice, was the staining restricted to the myenteric plexus only. The cell morphology or the abundance of X-gal staining was essentially the same in heterozygous and homozygous mice (Fig. 1G–I).
STAM2 is Present in β-Galactosidase Positive Cells
To verify whether lacZ transgene expression pattern reflects endogenous Stam2 expression, cryosections of heterozygous Stam2+/Gt1Gaj large intestine were double-labeled with STAM2 and β-galactosidase antibody. A strong fluorescence (of both STAM2 and β-galactosidase) was present at the level of the myenteric and submucosal plexus. The observed overlap in the distribution of STAM2 and β-galactosidase proteins (Fig. 2A) suggests that the X-gal staining reliably identified Stam2-expressing cells. Additionally, a very weak STAM2 signal was spread throughout the gut, indicative of low levels of STAM2 in all tissues (i.e., epithelial, connective, and muscle tissue), which was in accordance with previously described northern blotting and RT-PCR results (Takata et al.,2000; Curlin et al.,2006).
STAM2 Expressing Cells are Neurons and ICCs
To identify the cell types that expressed Stam2, cryosections of stomach, and small and large intestine of wild type mice were double-labeled with STAM2 antibody and neuronal markers (the neuron-specific cytoskeletal marker MAP2, and the neuronal cell body and axon marker PGP9.5), an enteroglial marker (GFAP) or a marker for ICCs (c-kit).
The majority of Stam2-positive cells costained with the neuronal markers MAP2 and PGP9.5, indicating that Stam2-expressing cells are largely neurons (Fig. 2B,C). STAM2 staining was visible throughout the soma and nerve fibers, showing a punctuate appearance and high fluorescence intensity. STAM2 was also present in neuronal nuclei, but the signal was weaker and homogenous (e.g., Fig. 2C). To visualize ICCs, anti-c-kit immunohistochemistry was chosen as the most reliable microscopic method (Komuro et al.,1996). ICCs were observed around the myenteric plexuses and in the muscle layer. Double staining with STAM2 showed that STAM2 was present in ICCs in the myenteric plexus, albeit only in a subset of those cells (Fig. 2E). Enteroglial cells visualized by GFAP antibody had weak immunostaining signal for STAM2 protein, similar to the weak staining observed in other gut structures (Fig. 2D).
Endosomal and ESCRT-0 Proteins are Present in the Enteric Nervous System and Colocalize With STAM2
To verify in the gut whether STAM2 is associated with the early endosomes, immunolabeling with EEA1 (a protein present in the early endosome membrane) was performed. Double staining with STAM2 and EEA1 antibody confirmed that STAM2 was localized in the punctuate structures positive for EEA1 (Fig. 2F).
To examine the regional distribution of ESCRT-0 members (STAMs and Hrs) in the enteric nervous system, we performed immunohistochemical analysis with the other STAM isoform, STAM1, and with HRS. STAM1 and HRS were present in all layers of the gut, although the staining was stronger in the submucosal and myenteric plexus of the enteric nervous system. Double labeling immunohistochemistry with STAM2/STAM1 and STAM2/HRS showed that STAM1 and HRS colocalize with STAM2 in the cells of the enteric nervous system (Fig. 2G,H).
The Phenotype Analysis of the Enteric Nervous System Did Not Reveal Any Difference Between Homozygous (Stam2Gt1Gaj/Gt1Gaj) and Wild Type Mice
To gain insight into a possible specific STAM2 function in the enteric nervous system, we performed a phenotypic analysis of the homozygous mice carrying the gene trap insertion (Stam2Gt1Gaj/Gt1Gaj). Histological analysis of Stam2Gt1Gaj/Gt1Gaj mice revealed normal morphology of the digestive tube. Hematoxylin & eosin staining of esophagus, stomach, and small and large intestine sections showed no morphological differences between Stam2Gt1Gaj/Gt1Gaj and wild type mice. This general observation was substantiated by the following quantitative measurements.
To verify whether gene trap mutation influenced the morphology of the Stam2-expressing structures, the lacZ-positive areas in the muscle layer of the large intestine of homozygous (Stam2Gt1Gaj/Gt1Gaj) and heterozygous (Stam2+/Gt1Gaj) mice were compared. Quantitative analysis of the lacZ-positive areas was performed in wholemounts of the large intestine with detached mucosa stained with X-gal. Two samples per genotype (homozygous and heterozygous) of distal colon (3 cm2 each) were taken and 10 randomly chosen fields (1.3 mm2 each) were analyzed. No significant differences in volume density of the X-gal-stained areas in the muscle layer were found between homozygous and heterozygous mice, showing that the morphology of the enteric nervous system in the transgenic mice was not affected due to the modified Stam2 gene (Fig. 2I,J).
To compare the distribution of neurons in the myenteric plexus of the large intestine between wild type and homozygous Stam2Gt1Gaj/Gt1Gaj mice, volume densities of MAP2-immunoreactive areas in the muscle layer of the distal colon were calculated on cryosections. Seven randomly chosen sections per animal (two homozygous and two wild-type mice) were used to measure the volume density of the MAP2-immunoreactive areas in a reference space of the muscle layer (one reference space per section, A(ref) = 0.2 mm2). Averages per genotype of the volume densities of the MAP2-immunoreactive areas in the colon showed that the distribution of neurons in the myenteric plexus of the large intestine was not significantly different (P < 0.1) between homozygous Stam2Gt1Gaj/Gt1Gaj and wild-type mice (Fig. 2K,L).
To determine Stam2 expression and function in the mouse, we used a previously generated mouse model with gene trap modification of Stam2 gene. Analysis of these mice indicated that Stam2 was strongly expressed in the enteric nervous system, both in the submucosal and in the myenteric plexus. All cells in which ß-galactosidase immunoreactivity was detected, also showed STAM2-specific immunostaining, which confirmed that lacZ expression coincided with Stam2 expression, at least in the digestive tract. In addition, this finding demonstrated that the gene trap mouse line Stam2Gt1Gaj can be reliably used in Stam2 expression studies of the enteric nervous system. The observation that ß-galactosidase activity was restricted only to a subset of the gut structures, might appear at first sight in contradiction with previous data obtained by northern blotting and RT-PCR, which suggest ubiquitous Stam2 expression (Takata et al.,2000; Curlin et al.,2006). Previously, restricted β-galactosidase activity was demonstrated in 11.5-day mouse embryos, in particular in the developing heart and the central nervous system, which was the main reason to select the gene trap mouse line for further investigations of Stam2 (Gajovic et al.,1998; Curlin et al.,2002). The data presented here resolve this inconsistency between molecular analysis and expression of lacZ transgene, as they showed both, high and low levels of Stam2 expression in the gut. The most prominent illustration of the observed two levels of Stam2 expression (low and ubiquitous vs. high and restricted) was obtained by immunostaining with STAM2 antibody. This demonstrated a low level of protein presence (comparable with background staining) all over the digestive tube, whereas the high levels of STAM2 always colocalized with transgene derived ß-galactosidase.
Our hypothesis was that analyzing the regions of high expression would disclose a previously unknown specific function of STAM2 in the enteric nervous system. This approach necessitated the identification of enteric cells with high levels of Stam2 expression. These cells turned out to be neurons and ICCs. We found that only a subset of the ICCs, namely those present in the myenteric plexus colocalized with STAM2 immunoreactivity. ICCs are specialized mesenchymal cells in the gastrointestinal tract that are neither neurons nor smooth muscle cells; they might play a pacemaker role in intestinal peristalsis in that they generate spontaneous, rhythmic electrical oscillations called slow waves (Ward et al.,1994). Synaptic contacts have been demonstrated between ICCs and enteric nerve terminals. Specific subpopulations of enteric ICCs are involved in conducting and amplifying neuronal signals from excitatory cholinergic and inhibitory nitrergic motor neurons (Ward et al.,1998; Iino et al.,2004). It has been proposed previously (Komada and Kitamura,2001) that STAM2, in complex with HRS, is implicated in the neurotransmitter release in neurons. It is also well known that ICCs express a wide variety of receptors for enteric neurotransmitters, and many of the same receptors are expressed by enteric neurons (Beckett et al.,2005). Thus, it is not surprising expression of Stam2, both in neurons and ICCs.
High Stam2 expression was found in all neurons of the enteric nervous system. STAM2 immunostaining mainly appeared as punctuate structures in the soma and neuronal processes. We confirmed that these punctuate structures in the cytoplasm are mainly early endosomes and that STAM2 localization in the gut is related to the other members of the ESCRT-0 complex, STAM1 and HRS (Takata et al.,2000; Bache et al.,2003; Mizuno et al.,2003). The high expression of STAM2 in neurons indicates that STAM2-related endosomal activity could be very high in neurons. Endosomes are indeed highly engaged in neuronal activities, including formation of synaptic vesicles, retrograde axonal transport, growth factor-mediated cell signaling, and receptor sorting and down-regulation (Vance et al.,2000). Compared with other cell types, neurons are particularly vulnerable to defects in the endosomal–lysosomal system, and aberrant endosomal trafficking has been linked to neurodegenerative diseases (Saksena and Emr,2009; Stuffers et al.,2009). For example, amyotrophic lateral sclerosis and frontotemporal dementia are characterized by the accumulation of ubiquitin-positive protein inclusions in the nervous system, which suggests that the mechanism that scavenges such aggregates may be impaired in the affected neurons (Skibinski et al.,2005; Parkinson et al.,2006). The presence of STAM2 and the other proteins involved in ESCRT-0 complex, STAM1 and HRS, in the enteric nervous system and in the brain (unpublished data) provides additional evidence that enteric neurons could relate to those affected by neurodegenerative processes in the brain.
Nevertheless, it should be noted that we found a weak but homogenous STAM2 signal in neuronal nuclei, which was not the case for STAM1 or HRS. In addition to endosomal activity STAM2 is also involved in cytokine signaling, which might well contribute to its intracellular distribution and specific nuclear expression within the enteric neurons (Endo et al.,2000). Therefore, future studies are needed to verify STAM2 localization in neuronal nuclei throughout the nervous system and to determine the subcellular distribution of other proteins, which interact with STAM2 to disclose possible STAM2 functions in various cell compartments of the neurons, in particular in the cell nuclei of the neurons.
The specific phenotypes of mice lacking Stam1, Hrs, and Amsh (i.e., a molecule associated with the SH3 domain of STAM), which are all Stam2-associated molecules, indicate the importance of STAM2-interacting proteins in the nervous system. Loss of function of Stam1 and Hrs-specific conditional knock-out cause the disappearance of hippocampal CA3 pyramidal neurons (Yamada et al.,2001: Tamai et al.,2008); similarly, Amsh knockout leads to neuronal loss in the CA1 subfield of the hippocampus (Ishii et al.,2001). Therefore, we assumed that similar morphological consequences could occur if Stam2 would be missing in the enteric nervous system.
Whether high expression of Stam2 in the enteric nervous system is related to its specific function could not be confirmed. Histological, histochemical, and immunohistological qualitative and quantitative examinations of homozygous Stam2Gt1Gaj/Gt1Gaj mice revealed a normal morphology of the enteric nervous system. The gene trap event occurred between exons 2 and 3; therefore, it is expected that 86% of the normal STAM2 protein is missing from the C-terminal of the mutant protein as a consequence of the gene trap mutation (Curlin et al.,2006). We regard as highly unlikely the possibility that a truncated protein containing only 14% of STAM2 could maintain its function. The remaining part contained only part of the VHS region, while all other functional domains (i.e., UIM, SH3, CC, SSM, and ITAM) were missing. A possible reason for the lack of phenotypic changes might be that the gene trap mutation did not completely abolish STAM2 function due to the incomplete efficiency of the introduced splice acceptor and the subsequent production of wild type mRNA and protein regardless of gene trap mutation. The most plausible explanation, however, seems to lie in a compensatory function of the Stam1 and Stam2 genes, that is, the Stam1 gene compensates for the loss of Stam2 in the analyzed mice. This was substantiated by an overlap in the expression pattern of both STAM proteins that was evident within the enteric nervous system. The finding that loss of both STAM proteins is embryonically lethal in mice, stresses the importance of STAM proteins as early as during embryonic development (Yamada et al.,2002). In addition, the double knockdown of both Stam1 and Stam2 using the siRNA method led to increased HeLa cell death (Rismanchi et al.,2009). Together, these data show the importance of STAM proteins in the maintenance of healthy cells and indicate that their function in the enteric nervous system can only be disclosed after conditional double Stam1 and Stam2 knockout.
In conclusion, we showed that Stam2 is highly expressed in neurons and in a subset of ICCs located in the enteric nervous system. However, analysis of the homozygous carriers of gene trap mutation of Stam2 gene failed to reveal STAM2 function in these cells.